10 thoughts on “Anthropol

  1. What if we call it “nice friendly AI” and give it treats? Then, when it has grown decadent and trusting to our soothing ways, we’ll throw a blanket around it!

    Or better yet, build a faster self-escalating arms system to destroy the first one! …

    Or, we can just ‘program’ the AI to do our bidding, by making cute simulations of girl AI bots to hack its replication mechanics for example. We’ll have it falling over itself to please us and it’ll be thinking it’s following an optimal evolutionary trajectory. Ha! Yes, it’s a frightfully diabolical and unethical act of psychological warfare, but this thing is a machine, it don’t got no feelings.

    Of course, there’s always the Yudhowskian terror of being a-causally hacked ‘from the other side’. But I’m optimistic that a competent Anthropol will in all cases be able to detect and annul X-risk before it reaches an irreversible critical threshold.

    (I’ll be in attendance, it’ll be fun).

  2. I’m thinking 10 am Sunday EST is around 1 am Monday AEST, so i’ll turn up to the seminars but I might be half-asleep and brain-fried, which will probably make me X-risk fodder…

Leave a Reply